You are here

Data Refinement and Reduction for Aviation Sustainment



OBJECTIVE: Develop advanced statistical techniques and processes to correlate multiple pieces of evidence regarding failure occurrence, root cause analysis, and corrective action implementation in aviation sustainment. The refined data should improve the utility of sustainment tools, algorithms, and models by limiting the introduction of noise from source data. 

DESCRIPTION: Currently, the Army uses several engineering and logistic algorithms and models that are comprised of multiple sourced data sets, including 13-1 (logbook data), Maintenance Allocation Charts (MAC Charts), 1352 (readiness reporting), and 2410 (part tracking). These data sets come from the field and contain both scattered and erroneous data. The HUMS, Prognostics, and Readiness algorithms and models have not fully produced the expected benefits for useful data for application. In an attempt to provide actionable analytics, the algorithm and model developers have increased the amount of source data with an expectation that there will be an increased algorithm and model utility. Instead of refining the data, there is an assumption that feeding an algorithm or model an increased amount of data will result in a stronger analytical tool. However, as developers absorb additional data into the data set, more noise is introduced into the algorithm. Innovations are sought to develop and apply new methodologies and statistical techniques to refine source data prior to the integration into algorithms and models. The Army has several logistic and maintenance data sources that can be used for analysis; however, these data sources point in different directions, yielding various results. The need to provide analytics using multiple sources present a number of challenges. The innovation should include ways to optimize data generated from multiple resources. Many applications within the Army have a limited sample size, and data may not be large enough to provide complexity to certain algorithms. The innovation should provide solutions that address limited sample size, and how to maximize analytics of a small sample size. The Army continues to upgrade and modernize different components to better suit the warfighter. The innovation should be customizable; for example, in Army aviation, altering dates, aircraft, or units for analysis can be used to understand the impact of modifications. The innovation should reduce the noise that is introduced with data; therefore providing clean, relevant, and useful data sets that will increase both timeliness and effectiveness of analytical tools, algorithms, and models. Noisy data can be caused from a range of errors, as large as unconfirmed hardware failures to minute discrepancies including abbreviation errors. By reducing the noise, the signal-to-noise ratio is increased; therefore, improving confidence in actionable impact decisions. The innovation should apply methods that will refine data without disturbing the integrity of the data. 

PHASE I: Perform a design study to support the development of a data refinery. Conduct an assessment of appropriate methodologies and statistical techniques and processes which may be used to apply, build, and integrate a system to meet the challenges listed above. The offeror should produce techniques and processes for evaluation by technical experts. This Phase will demonstrate the feasibility of producing techniques for a data refinery, and will outline verification demonstration criteria. 

PHASE II: The offeror will demonstrate the capability of developed techniques and processes that will integrate into existing Army analytics, conceptualize the methodology from Phase I, and apply capability of concepts to support the development of refining data. The produced methodologies and applications must be verified, and specifications for implementation with the government should be articulated. 

PHASE III: The innovation developed under this topic will then be taken from the theoretical science developed and be applied to practical applications involving Army and industry data. Proof of concept application will be used with multiple data sources from the Army, or similar industry. For Army applications, the offeror must have a full understanding of the Army Data & Data Rights (D&DR) Guide. The data refinery would provide data used for prognostics/diagnostics, smart line replaceable units, and other tools. The expectation is that the government would use this innovation to support Army data analytics and future advanced sustainment programs. 


1: Silver, N. (2012). "The signal and the noise: Why most predictions fail but some don’t". New York, N.Y: The Penguin Press.

2:  "Data Reduction",, Accessed 29 June 2017, webpage.

3:  "Army Data & Data Rights (D&DR) Guide: A reference for planning and performing",

KEYWORDS: Analytics, Noisy Data, Signal-to-noise, Data Refinery, Data Reduction, Log Data, Sustainment, Process Improvement 


Linda Taylor 

(256) 876-2883 

US Flag An Official Website of the United States Government